This is the accepted version of the article: Liu, Lei; Estiarte, Marc; Peñuelas, Josep. «Soil moisture as the key factor of atmospheric CH4 uptake in forest soils under environmental change». Geoderma, Vol. 355 (Dec. 2019), art. 113920. DOI 10.1016/j.geoderma.2019.113920 This version is avaible at https://ddd.uab.cat/record/218215 under the terms of the license 1 TITLE: Soil moisture as the key factor of atmospheric CH4 uptake in forest soils under 2 environmental change 3 4 RUNNING TITLE: Atmospheric CH4 uptake by forest soils 5 6 AUTHORS: Lei Liu1, 2, Marc Estiarte2, 3, Josep Peñuelas2, 3 7 8 1. Institute of Ecology, Jiangsu Key Laboratory of Agricultural Meteorology, Nanjing 9 University of Information Science & Technology, Nanjing 210044, China. 10 2. CREAF, E08193, Cerdanyola del Vallès, Catalonia, Spain. 11 3. CSIC, Global Ecology Unit, CREAF-CSIC-UAB, E08193 Bellaterra, Catalonia, 12 Spain. 13 14 Corresponding author: Lei Liu 15 Email: liulei406@gmail.com 16 KEY WORDS: CH4 uptake; forest; Environmental change; Elevated CO2; N 17 enrichment; Drought; Warming 18 1 19 Abstract 20 Methane (CH4) is an important anthropogenic greenhouse gas that can be produced and 21 consumed by microorganisms in soils. We present a meta-analysis of the potential 22 effects of environmental change on CH4 uptake by forest soils. Such effects have not 23 been reliably estimated even though aerobic methanotrophs in forest soils are the largest 24 biological sink for atmospheric CH4. Differences in the annual rate of CH4 uptake 25 between forests are likely caused by differences in vegetation, microbial communities, 26 and the physic-chemistry of soil environments, but we found no clear different patterns 27 at annual scale among tropical, temperate, and boreal forests. The meta-analysis 28 indicated that the rates of CH4 uptake in forest ecosystems were significantly decreased 29 under elevated CO2 and N enrichment, but the rates increased under drought. The 30 effects of warming on the rates of CH4 uptake were inconsistent in forest soils, and the 31 response ratio accordingly suggested that a warmer climate would have no significant 32 effect on the rate of CH4 uptake. The seasonality of CH4 uptake in natural forest soils 33 and the clear results of the drought experiments evidence the importance of soil 34 moisture. However, our linear model did not unravel a clear negative effect of climatic 35 water surplus nor mean annual precipitation on soil CH4 uptake. Therefore, process- 36 based and ecosystem-specific models of CH4 flux are also warranted for predicting the 37 responses of ecosystemic CH4 fluxes to climate change. 38 2 39 1. Introduction 40 Methane (CH4) is an important greenhouse gas with a warming potential 25 times 41 greater than that of CO2 and is responsible for about 20% of the realized global warming 42 (IPCC, 2007). Atmospheric CH4 concentration has been increasing from the pre- 43 industrial value of around 715 ppb to the current value of near 1800 ppb (Heimann, 44 2011). The global annual rate of the increase of atmospheric CH4 caused by the 45 imbalance between sources and sinks, decreased from an average of 3.3 ppb y-1 in the 46 1980s to 1.3 ppb y-1 in the 2000s, although the rate began to increase again in 2007 47 (Kirschke et al., 2013). The decadal and inter-annual variation of the rate of increase is 48 not yet fully understood. Two alternative causes have been suggested to explain the 49 reduction over the last three decades (Heimann, 2011): one suggests a decrease in fossil- 50 fuel emissions (Aydin et al., 2011), and the other suggests a decrease in emissions from 51 rice cultivation in Asia due to higher fertilizer application and reduction in water use 52 (Kai et al., 2011). 53 Most of the atmospheric CH4 is oxidized in the troposphere by chemical reactions with 54 hydroxyl radicals (OH), which comprises approximately 90% of the global sinks 55 (Schlesinger and Bernhardt, 2013). The second largest global sink of atmospheric CH4 56 is the consumption by aerated soils. The amount of CH4 oxidized by soil methane- 57 oxidizing bacteria (MOB) was estimated to be between 26 and 42 Tg y-1 for 2000-2009 58 (Kirschke et al., 2013). Forest soils represent approximately 50% of this sink and coarse 59 forest soils have the highest rates of CH4 uptake (Dutaur and Verchot, 2007). 60 The main factor regulating the CH4 uptake capacity of soils is the diffusion rate of gases 3 61 that regulates the availability of CH4 to MOB across the soil profile. A number of soil 62 characteristics, such as texture, structure, moisture content, temperature, and mineral- 63 nitrogen (N) content, are important to CH4 uptake. The diffusion of gases in soils 64 structurally depends on the soil texture and degree of compaction (Castaldi and Fierro, 65 2005) because they affect either the diffusion of gas through the soil or the activity and 66 size of the soil microbial populations involved in CH4 metabolism (Castaldi and Fierro, 67 2005; King, 1997; Lin et al., 2015; Price et al., 2003; Verchot et al., 2000). Although 68 the relationship between CH4 uptake and soil moisture was identified decades ago 69 (Bowden, 1998), uncertainty remains as to how future climate change will affect the 70 CH4 uptake in soils of forest ecosystems and their role in the global CH4 cycle. 71 Several components of environmental change, such as altered precipitation, warming, 72 rising atmospheric CO2 concentration and increased atmospheric N deposition, can 73 potentially alter the soil properties or biology that control the uptake of CH4 in forest 74 soils and determine the size of their atmospheric CH4 sink (Blankinship et al., 2010). 75 The effect of environmental change on CH4 uptake by forest soils has not been reliably 76 estimated, although forests have been responsible for most of the terrestrial uptake of 77 atmospheric CH4. We reviewed the literature with three main aims: (1) to synthesize 78 the information available on the rates of CH4 uptake in forest soils in different biomes, 79 (2) to evaluate the response of CH4 uptake to components of environmental change and 80 to summarize the current state of our understanding of the mechanisms underlying the 81 responses, and (3) to test for across-sites controls by climate and water balance in CH4 82 uptake in forest soils. We also identify topics requiring further study. 83 4 84 2. Materials and methods 85 We gathered data on CH4 uptake rates in the forests and under different treatments to 86 thereafter conduct a meta-analysis and accomplish the first two aims. For the third aim 87 we also gathered data on climate and soil moisture from world databases and used them 88 to model their relations with the CH4 uptake by forest soils. 89 2.1. Data source. 90 We systematically searched all peer-reviewed journal articles and those that 91 investigated CH4 uptake in forest ecosystems. The literature search was done through 92 the Web of Science seeking for the keywords “methane or CH4” and “forest”. We 93 reviewed all the found articles and selected those that met the following two criterias: 94 (1) the study was conducted in situ field measurements for several months; (2) the CH4 95 uptake could be extracted directly from the texts, tables, and figures. When several 96 publications include data from the same locations we obtained the data as annual 97 average. When one publication includes several experiments under different abiotic 98 conditions, such as different locations, tree species, or stand ages, we considered them 99 different observations. In total we found 134 datasets for forest soils at 134 sites from 100 110 papers in 26 countries (Table S1). Most of the studies used static chambers, but 101 some used stable isotopes. The mean annual CH4 uptake rates published in the primary 102 or secondary literature were used when provided and when annual averages were not 103 provided they were calculated based on figures. We used Plot Digitizer version online 104 to digitally extract data from figures when the results were graphically reported. CH4 5 105 uptake rates were standardized to kg CH4 ha-1 y-1. The study sites comprise tropical 106 forests, coniferous-, deciduous-, or mixed-temperate forests and boreal forests (Fig. 1). 107 For each site identified by the latitude and longitude coordinates, we extracted monthly 108 average temperature and precipitation from the WorldClim database 109 (http://www.worldclim.org/) with a spatial resolution of around 1 km at the equator, and 110 aggregated them to mean annual temperature (MAT) and mean annual precipitation 111 (MAP). Sites were climatically characterized by MAT, MAP, temperature and 112 precipitation during the summer month (Tsum and Psum) and potential 113 evapotranspiration (PET). We defined summer month as July in the Northern 114 hemisphere and January in the Southern hemisphere. PET was estimated on a monthly 115 basis from monthly average temperature and latitude with the function “thornthwaite” 116 from the R package:spei. 117 In addition to CH4 uptake, we estimated soil water-holding capacity (SWHC) from soil 118 moisture data obtained from the Soil Moisture and Ocean Salinity (SMOS) database 119 (https://smos-ds-02.eo.esa.int/oads/access/). For every site and from June 2010 to June 120 2017, we gathered the products that comprise soil moisture measurements geo-located 121 in an equal-area grid system ISEA 4H9 as percent of water-filled soil volume in “SMOS 122 level 1 and 2 Science data” collection and with a spatial resolution in the range of 30- 123 50 km. We assumed that during the period 2010-2017 maximum and minimum soil 124 moisture levels were reached and we proceeded to obtain maximum and minimum soil 125 water content after removing the 5% extremely low or high values. For each site, we 126 calculated SWHC as maximum minus minimum soil water content and extended it as 6 127 the soil was 50 cm deep. In total, SWHC was available for 124 sites. Afterwards, soil 128 moisture data from satellite was standardized to the site amplitude in the historical 129 record and expressed as a percent of the estimated site SWHC which we coined as 130 standardized soil moisture (SSM). The mean annual SSM (MASSM), summer SSM 131 (SSMsum) and winter SSM (SSMwin) were also included in the database. We defined 132 summer month as July or January and winter month as January or July respectively for 133 the Northern or the Southern hemisphere. 134 For each site, we used Stephenson’s bucket (1990) approach to adjust a climatic water 135 balance using average monthly values of PET, precipitation and site SWHC to estimate 136 monthly values of actual evapotranspiration (AET), water deficit (WD), water surplus 137 (WS) and mean annual soil water content (SWC) (details for the estimation of water 138 balance variables are in the Appendix.). Water balance variables indicate how much 139 energy and water are available at the same time (AET), how much evaporative demand 140 is not met by available water (WD) and how much water is unusable surplus (WS) 141 (Stephenson, 1990). 142 2.2 Meta-analysis. 143 To examine the effects of environmental-change components on the CH4 uptake rate in 144 forest ecosystems, among the 134 datasets gathered previously, we selected the datasets 145 including experimental treatments simulating the following components of the 146 environmental change: elevated atmospheric CO2, warming, drought or water addition, 147 and N enrichment. To increase the number of datasets, we also included studies with a 148 reduced number of field measurements as well as data from laboratory incubations with 7 149 warming treatment or with N addition. Details of the studies are summarized in 150 supplementary Table S2. We calculated the response ratios from each study as described 151 by Hedges et al. (1999). Briefly, the natural-log response ratio (lnRR) was calculated 152 as: 153 ln (Xi/Xn) = lnXi  lnXn 154 where Xi and Xn are the values of each observation in the treatment and corresponding 155 control plots, respectively. The sampling variance for each lnRR was calculated as: 156 ln[(1/ni) × (Si/Xi)2 + (1/nn) × (Sn/Xn)2] 157 where ni and nn, Si and Sn, and Xi, and Xn are the treatment and control sample sizes, 158 standard deviations, and mean responses, respectively. The natural-log response ratios 159 (from here onwards simplified as response ratios) were determined by specifying 160 studies as random factors using the rma model in the R metafor package. The effects 161 on CH4 uptake rates and the differences between the treatment and control plots were 162 considered significant if the 95% confidence interval (CI) of lnRR did not overlap zero. 163 All statistical analyses were performed in RStudio 3.1.2 (R Core Team 2015) using the 164 R package metafor 1.92. 165 2.3 Linear modelling of CH4 uptake 166 We built linear models to explain forest annual CH4 uptake at site level from climatic 167 variables, standardized soil moisture variables and derived water balance variables. 168 Water balance variables are a lineal combination of MAP and PET (e.g PET = 169 AET+WD; MAP=AET+WS), so MAP and PET were excluded when water balance 170 variables were allowed to enter the models. 8 171 Model selection was performed using procedures based on AIC implemented in the 172 Mumin package in R environment Partial residual plots of the models were obtained 173 using the visreg package to evaluate the effect of each variable on the model. 9 174 3. Results 175 3.1 CH4 uptake in forest soils 176 The mean uptake rate for all studies was 4 kg CH4 ha-1 y-1, with bootstrapped 95% CIs 177 between 1.77 and 5.85. Ninety percent of a total of 134 observations were between 178 0.119 and 15.82 kg CH4 ha-1 y-1. The highest reported uptake rate was 40.52 kg CH4 179 ha-1 y-1 in a tropical forest in India. The lowest reported rate was -120 kg CH4 ha-1 y-1 180 in a boreal forest, but this was an unusually low value. Only five of the observations 181 were net CH4 sources, representing 3.7% of all observations, whereas the remaining 182 129 observations were net sinks, with 13.4% of the values between 0 and 1 kg CH4 ha- 1 183 y-1 and 81.3% above 1 kg CH4 ha-1 y-1. No clear differences among forest types were 184 detected. Mean and bootstrapped 95% CIs were 4.08 (2.42, 6.89), 5.1 (3.96, 6.34), and 185 -2.74 (-21.5, 8.16) for tropical, temperate, and boreal forests, respectively, with the low 186 rates of the boreal forests driven by a single site with an extremely low rate, without 187 which the mean increased to 5.64 (2.61, 9.22). Median rates were 2.82, 3.32, and 2.62 188 kg CH4 ha-1 y-1 for tropical, temperate, and boreal forests respectively (Fig. 1). 189 190 3.2 Soil CH4 uptake across sites 191 After removing two sites with the lowest and the highest CH4 uptake, the database used 192 for the modelling included 132 sites and included climatic data for all of them. The best 193 linear model fitted to the soil CH4 uptake in the 132 sites and including only climate as 194 explanatory variables was MAT*MAP (adj R2= 0.10). There was almost no variance 195 explained (adj R2 = 0.02) if the interaction was not included. 10 196 We used the subset of 107 sites having SMOS data for all the twelve months to include 197 the SMOS derived soil moisture. (Table S3). The addition of SMOS derived soil 198 moisture variables (MASSM, SSMsum SSMwin) did not change the model selection 199 and the MAP*MAT model behaved similarly with the subset including 107 sites (adj 200 R2= 0.09). When water balance variables substituted MAP and PET the best model was 201 MAT*WS (adj. R2 = 0.12). Including Tsum yielded a model with lower AICc (and adj 202 R2 = 0.14) but Tsum was discarded because it was non-significant after model averaging. 203 Partial residual plots (Fig. 2) show the interaction effect of MAT and WS on soil CH4 204 uptake. Higher WS occurs only at the warmer sites of the dataset and low and mild WS 205 occurred at forests within the three temperature intervals. Soil CH4 uptake did not 206 respond to WS across the warmer sites, which presented uptake values in the low range. 207 In the milder and colder sites CH4 uptake tended to increase from low to mild WS sites. 208 A slight effect of MAT was present only at the wetter sites. 209 210 3.3 Alteration of atmospheric CH4 uptake in forest soils by environmental-change 211 components 212 The meta-analysis of 10 experiments indicated that the CH4 uptake by soils in forest 213 ecosystems significantly increased an average of 150% under drought conditions. In 214 contrast, the meta-analysis of 7 experiments showed that elevated CO2 significantly 215 decreased an average of 33% the CH4 uptake in temperate forests, the only forest type 216 where data on CH4 uptake on CO2-enrichment experiments have been reported. 217 Similarly, the meta-analysis of 29 experiments showed that N-enrichment significantly 11 218 decreased CH4 uptake rates an average of 36% (Fig. 3). Our meta-analysis of 6 219 experiments did not identify a clear effect of ecosystem warming on CH4 uptake by 220 forest soils (Fig. 3). 221 12 222 4. Discussion 223 Previous work with a dataset including all terrestrial ecosystems found that MAP and 224 MAT explained about 3% and 2% of the global CH4 uptake variation (Dutaur and 225 Verchot 2007). Our modelling for forest ecosystems explained a bit more of the 226 variance but neither showed a strong climatic control across forests in spite of 227 differences in vegetation. Different tree species likely produce litter of different quality 228 and soils differing in organic matter content and in chemical, physical and biological 229 characteristics that are important for CH4 uptake (Barrena et al., 2013; Borken and 230 Beese, 2006). Lower rates of CH4 uptake were found in coniferous than in deciduous 231 forest soils (Borken et al., 2003), suggesting the occurrence of differences among broad 232 forest types. Differences in CH4 uptake are also related to differences in the structure 233 and function of the soil microbial community (Aronson et al., 2013; Nazaries et al., 234 2011) and communities may differ among forests. For instance, lower diversity and 235 abundance of methane-oxidizing bacteria in spruce than in beech forest soils 236 (Degelmann et al. 2010) suggests that tree species may influence the activity of soil 237 methane-oxidizers. Nevertheless, the lack of clear differences in CH4 uptake among 238 forest types indicates low climatic control. 239 Beyond climate, at the biome scale soil-textural class is an important determinant of 240 soil CH4 fluxes (Verchot et al., 2000), with coarse and medium-textured soils 241 consuming more CH4 than fine-textured soils (Dutaur and Verchot, 2007). Soil texture 242 is a local characteristic and the proportion of mineral particles is not going to be altered 243 by environmental change. Soil aeration depends on the soil water content that fills the 13 244 networks of small pores and impedes the transport of gases (Hartmann et al., 2011; 245 Hiltbrunner et al., 2012) and is most likely going to be altered by climate change. 246 Similarly soil temperature and nutrient content will also be altered. Actually, moisture 247 and temperature are temporally variable under natural conditions and are underlying 248 the seasonality in CH4 uptake by forest soils. For example, some tropical forests shift 249 between sources in wet seasons to sinks in dry seasons or increase the sink strength 250 from wet to dry seasons (Teh et al., 2014). Temperate and boreal forests may also 251 release CH4 under wet or water-saturated conditions (Gundersen et al., 2012). Similarly, 252 differences in CH4 uptake among years in the same forest are also dependent on 253 differences in precipitation (Matson et al., 2009). 254 The meta-analysis of drought experiments confirmed short term effects of moisture on 255 soil CH4 uptake. Increases in CH4 uptake under experimental drought have been 256 described in tropical, temperate and boreal forests (Billings et al., 2000; Borken et al., 257 2006; Davidson et al., 2008, 2004). Most effects of drought are likely due to improved 258 diffusion of atmospheric gases into the soil, although very low moisture can have direct 259 effects on microbial activity. A lower diffusion alters the CH4 source-sink balance of 260 soils because reduces the supply of atmospheric CH4 into the soil leading to substrate 261 limitation of the activity of methanotrophic organisms (Blankinship et al., 2010), and 262 reduces the supply of O2, which may increase methanogenesis (Borken et al., 2006). 263 When soil moisture is too low, gases diffuse without restriction but microbes are 264 physiologically stressed and its activity, including methanotrophs, is reduced, which 265 reduces the CH4 uptake (Price et al., 2004). The stimulation of methanotrophic activity 14 266 after rains in deserts and semiarid regions provides evidence of the stress imposed on 267 methanotrophs in very dry soils (McLain and Martens, 2005). Therefore, the sensitivity 268 of soil CH4 uptake to soil moisture can be described by a parabolic curve reflecting that 269 CH4 uptake is limited at very low moisture by biological activity and is limited by CH4 270 diffusion at high moisture (Fest et al., 2017). 271 The seasonality of CH4 uptake in forest soils and the clear results of the drought 272 experiments evidence the importance of soil moisture. Recent studies show that a 273 decline in CH4 uptake at a global scale coincides with increases in precipitation in forest 274 soils (Ni and Groffman, 2018; Yu et al., 2017). Accordingly, we expected wetter soils, 275 less aeration and lower CH4 uptake at high MAP and, specially, at high climatic WS 276 (note that MAP and WS are highly correlated). However, satellite-derived soil moisture 277 and climatic water balance variables did not increase much the explanation of the 278 variance in CH4 uptake. The modelling did not unravel a clear negative effect of 279 climatic WS nor MAP on CH4 uptake. Certainly, the uncertainties introduced by the 280 type of data used could partly account for the poor fit. The spatial resolution of the 281 satellite data in the range of 30-50 km can introduce large errors because soil 282 characteristics can vary at a much smaller scale. Moreover, precipitation may also vary 283 at relatively small scales. 284 The possible effects of the other environmental change components addressed by our 285 meta-analysis on soil CH4 uptake were less evident than for drought. Meta-analysis did 286 not reveal warming effects on soil CH4 uptake, although the low number of reviewed 287 experiments prevents considering it a definitive result. We must keep in mind that 15 288 warming in the field often leads not only to warmer but also to drier soils (Luo et al., 289 2013), which may increase gas diffusion and CH4 uptake rates (Price et al., 2003). 290 Warming may also accelerate soil N mineralization, leading to higher NH4+ 291 concentrations that might suppress CH4 uptake (Karbin et al., 2015). The opposite 292 effects of improved gas diffusion and of increased soil NH4+ (Lüke and Frenzel, 2011) 293 may explain that warming in field experiments has shown scarce effects on CH4 uptake 294 (Karbin et al., 2015; Price et al., 2004). 295 The meta-analysis provided more clear evidence of the effects of elevated atmospheric 296 CO2 despite the number of experiments was as low as for warming. Dijkstra et al. (2012) 297 reported that elevated CO2 tended to increase CH4 emissions in wetlands, peat lands, 298 and rice paddy fields, but the effects were highly variable in upland soils. We found an 299 average of 33% decrease (Fig. 3) in the CH4 uptake by soils of temperate forests, the 300 only forest type where CO2-enrichment experiments have been reported. The decrease 301 was very clear at some sites (Dubbs and Whalen, 2010; Phillips et al., 2001) and was 302 associated to increases in soil moisture. The lower CH4 uptake could be due to a reduced 303 CH4 diffusion into moister soils, although increases in CH4 production by methanogens 304 were also likely (Dubbs and Whalen, 2010; Phillips et al., 2001). The reduction in CH4 305 uptake under elevated CO2 is relevant because 20% of the world’s forests are temperate 306 (Pan et al., 2013). However, tropical forests represent 51% and boreal forests 29% of 307 the world’s forests so research on the response of their soil CH4 uptake to atmospheric 308 CO2 increase is warranted. 309 Anthropogenic N enrichment has a suite of detrimental effects on ecosystem services 16 310 and the meta-analysis confirmed the decrease in CH4 uptake rates. Some experiments 311 reported a very clear inhibitory effect of the fertilization with N on CH4 uptake 312 (Adamsen and King, 1993; Fender et al., 2012; Steinkamp et al., 2001; Wang and Ineson, 313 2003) although a synthesis of studies in non-wetland ecosystems reported dose- 314 dependent effects, with smaller N enrichments tending to stimulate soil CH4 uptake and 315 larger ones tending to inhibit it (Aronson and Helliker, 2010). These opposite N dose- 316 dependent effects suggest that the historical N status of soils is the most important 317 predictor of the response of CH4 uptake to future N inputs (Aronson and Helliker, 2010). 318 Increase in soil NH4+ may reduce soil CH4 uptake because NH4+ competes with CH4 at 319 the reaction site of the enzyme methane monooxygenase, the first step of the CH4 320 oxidation pathway (Bodelier and Laanbroek, 2004). The non-competitive inhibition of 321 CH4 uptake by NO3- have been attributed to the increase in NH4+ concentrations caused 322 by NO3- (Fender et al., 2012). Furthermore, nitrate (NO3-) has also been reported to 323 have direct inhibitory effects on CH4 uptake (Fender et al., 2012; Mochizuki et al., 2012; 324 Wang and Ineson, 2003). 325 We reviewed individual components of global environmental change, but it is still 326 unknown whether the effects of multiple components are additive, cancel each other, or 327 synergistically increase the individual effects. Ambus and Robertson (1999) reported 328 that elevated CO2 reduced CH4 uptake only when no N was deposited. Another study 329 found no interaction between altered precipitation and warming in four ecosystems 330 along a 50-km climatic gradient from warm and dry to cold and wet, although the 331 authors suggested that a wet and warm climate would cause the largest reduction in 17 332 terrestrial CH4 uptake (Blankinship et al., 2010). The lack of interactive effects from 333 these experiments may be due to inadequate statistical power or because the time 334 required by the interactive effects to appear was longer than the duration of the 335 experiments (Norby and Luo, 2004). Longer field experiments are therefore needed to 336 unravel the interactive effects of climate change on soil CH4 fluxes. 337 Moreover, modeling complements field experiments, thus overcoming the difficulties 338 associated with long-term studies and identifying important interactive effects among 339 multiple factors of climate change on ecosystemic processes (Luo et al., 2008). For 340 example, the Dynamic Land Ecosystem Model, a process-based model, unraveled that 341 the interaction between environmental-change components (including climatic 342 variability, N deposition, elevated levels of atmospheric CO2 and application of N 343 fertilizer) led to a decrease in CH4 uptake over the last three decades in North America 344 (Xu et al., 2010). 345 The drought experiments show the key position of soil moisture in the short-term 346 control of CH4 uptake in forest soils. The weak control of climatic (i.e. long term) 347 precipitation on soil CH4 uptake across sites contrasts notoriously with the clear short- 348 term effects of experimental reduction in precipitation. It suggests that despite the 349 immediate effects of weather, the CH4 consumption is determined in the long term, after 350 texture is considered, by characteristics derived from vegetation, soil nutrients and 351 microbial communities that are weakly controlled by climate but strongly by local 352 conditions. The importance of the structural properties of ecosystems on soil CH4 353 uptake is evidenced when changing the tree-species or after transformation of forests 18 354 into grasslands (Hiltbrunner et al., 2012; Smith et al., 2000). Despite the absence of 355 evidences of warming effects, we cannot discard alterations in the CH4 uptake if 356 warming causes structural changes in the long term. The possible long term effects 357 mediated by soil and vegetation changes challenge the modelling and the prediction of 358 CH4 consumption and an effort is needed to disentangle which are the vegetation- 359 dependant soil traits governing CH4 uptake rates. 360 5. Conclusions and outlook 361 Our in-depth review of the effects of particular environmental-change components on 362 CH4 uptake in forest soils reveals that the rates were significantly reduced by elevated 363 CO2 and N enrichment, that were increased by drought and that were not consistently 364 altered by warming. Very few studies, however, have analyzed the interactive effects of 365 multiple environmental-change components due to the challenge of the complexity of 366 forests. More field experiments are, therefore, required to expand our knowledge of the 367 impacts of these multiple factors on CH4 uptake and CH4-cycling microbial 368 communities. The poor relation between climatic variables and CH4 uptake unveiled by 369 linear modelling may be influenced by uncertainties of the data sources, but it also 370 indicates that process-based and ecosystem-specific models of CH4 flux are necessary 371 for predicting the response of ecosystemic CH4 fluxes to climate change. 372 The studies on CH4 uptake in natural soils have ignored the environmental importance 373 of phosphorus, which is becoming unbalanced due to increased C and N availabilities 374 (Peñuelas et al., 2013). It is urgent to incorporate P on the agenda to provide data on its 375 possible significance for the CH4 uptake and for the abundance, activity and structure 19 376 of methanotrophic communities. Attention should also be paid to the indirect effect of 377 climate change on plant communities and the decomposition of litterfall in forest 378 ecosystems because potential changes can directly affect factors important for CH4 379 uptake in forest ecosystems such as moisture content, temperature and gas diffusion in 380 soils. 20 381 Acknowledgements 382 This research was supported by the international postdoctoral exchange fellowship 383 program by the office of China Postdoctoral Council (No. 20140051) and by the 384 European Research Council Synergy grant ERC-2013-SyG610028-IMBALANCE-P, 385 the Spanish Government project CGL2016-79835-P, and the Catalan Government grant 386 SGR 2017-1005 387 21 388 References 389 Adamsen, A.P.S., King, G.M., 1993. Methane consumption in temperate and subarctic 390 forest soils: Rates, vertical zonation, and responses to water and nitrogen. Applied 391 and Environmental Microbiology 59, 485–490. 392 Ambus, P., Robertson, G.P., 1999. Fluxes of CH4 and N2O in aspen stands grown under 393 amnient and twice-ambient CO2. Plant and soil 209, 1–8. 394 https://doi.org/10.1016/j.agee.2011.12.016 395 Aronson, A.E.L., Helliker, B.R., 2010. Methane flux in non-wetland soils in response 396 to nitrogen addition : a meta-analysis Linked references are available on JSTOR 397 for this article : Methane flux in non-wetland soils in response to nitrogen addition : 398 a meta-analysis. Ecology 91, 3242–3251. https://doi.org/10.2307/20788157 399 Aronson, E.L., Allison, S.D., Helliker, B.R., 2013. Environmental impacts on the 400 diversity of methane-cycling microbes and their resultant function. Frontiers in 401 Microbiology 4, 1–15. https://doi.org/10.3389/fmicb.2013.00225 402 Aydin, M., Verhulst, K.R., Saltzman, E.S., Battle, M.O., Montzka, S.A., Blake, D.R., 403 Tang, Q., Prather, M.J., 2011. Recent decreases in fossil-fuel emissions of ethane 404 and methane derived from firn air. Nature 476, 198–201. 405 https://doi.org/10.1038/nature10352 406 Barrena, I., Menéndez, S., Duñabeitia, M., Merino, P., Florian Stange, C., Spott, O., 407 González-Murua, C., Estavillo, J.M., 2013. Greenhouse gas fluxes (CO2, N2O and 408 CH4) from forest soils in the Basque Country: Comparison of different tree species 409 and growth stages. Forest Ecology and Management 310, 600–611. 22 410 https://doi.org/10.1016/j.foreco.2013.08.065 411 Billings, S.A., Richter, D.D., Yarie, J., 2000. Sensitivity of soil methane fluxes to 412 reduced precipitation in boreal forest soils. Soil Biology & Biochemistry 32, 413 1431–1441. 414 Blankinship, J.C., Brown, J.R., Dijkstra, P., Allwright, M.C., Hungate, B.A., 2010. 415 Response of Terrestrial CH4 Uptake to Interactive Changes in Precipitation and 416 Temperature Along a Climatic Gradient. Ecosystems 13, 1157–1170. 417 https://doi.org/10.1007/s10021-010-9391-9 418 Bodelier, P.L.E., Laanbroek, H.J., 2004. Nitrogen as a regulatory factor of methane 419 oxidation in soils and sediments. FEMS Microbiology Ecology 47, 265–277. 420 https://doi.org/10.1016/S0168-6496(03)00304-0 421 Borken, W., Beese, F., 2006. Methane and nitrous oxide fluxes of soils in pure and 422 mixed stands of European beech and Norway spruce. European Journal of Soil 423 Science 57, 617–625. https://doi.org/10.1111/j.1365-2389.2006.00752.x 424 Borken, W., Davidson, E.A., Savage, K., Sundquist, E.T., Steudler, P., 2006. Effect of 425 summer throughfall exclusion, summer drought, and winter snow cover on 426 methane fluxes in a temperate forest soil. Soil Biology and Biochemistry 38, 427 1388–1395. https://doi.org/10.1016/j.soilbio.2005.10.011 428 Borken, W., Xu, Y.J., Beese, F., 2003. Conversion of hardwood forests to spruce and 429 pine plantations strongly reduced soil methane sink in Germany. Global Change 430 Biology 9, 956–966. https://doi.org/10.1046/j.1365-2486.2003.00631.x 431 Bowden, K.M.N. and G.M.R., 1998. Carbon dioxide and methane fluxes by a forest 23 432 soil under laboratory-controlled moisture and temperature conditions. Soil 433 Biology & Biochemistry 30, 1591–1597. 434 Castaldi, S., Fierro, A., 2005. Soil-atmosphere methane exchange in undisturbed and 435 burned Mediterranean shrubland of southern Italy. Ecosystems 8, 182–190. 436 https://doi.org/10.1007/s10021-004-0093-z 437 Davidson, E.A., Ishida, F.Y., Nepstad, D.C., 2004. Effects of an experimental drought 438 on soil emissions of carbon dioxide, methane, nitrous oxide, and nitric oxide in a 439 moist tropical forest. Global Change Biology 10, 718–730. 440 https://doi.org/10.1111/j.1529-8817.2003.00762.x 441 Davidson, E.A., Nepstad, D.C., Ishida, F.Y., Brando, P.M., 2008. Effects of an 442 experimental drought and recovery on soil emissions of carbon dioxide, methane, 443 nitrous oxide, and nitric oxide in a moist tropical forest. Global Change Biology 444 14, 2582–2590. https://doi.org/10.1111/j.1365-2486.2008.01694.x 445 Dijkstra, F.A., Prior, S.A., Runion, G.B., Torbert, H.A., Tian, H., Lu, C., Venterea, R.T., 446 2012. Effects of elevated carbon dioxide and increased temperature on methane 447 and nitrous oxide fluxes: Evidence from field experiments. Frontiers in Ecology 448 and the Environment 10, 520–527. https://doi.org/10.1890/120059 449 Dubbs, L.L., Whalen, S.C., 2010. Reduced net atmospheric CH4 consumption is a 450 sustained response to elevated CO2 in a temperate forest. Biology and Fertility of 451 Soils 46, 597–606. https://doi.org/10.1007/s00374-010-0467-7 452 Dutaur, L., Verchot, L. V., 2007. A global inventory of the soil CH4 sink. Global 453 Biogeochemical Cycles 21, 1–9. https://doi.org/10.1029/2006GB002734 24 454 Fender, A.C., Pfeiffer, B., Gansert, D., Leuschner, C., Daniel, R., Jungkunst, H.F., 2012. 455 The inhibiting effect of nitrate fertilisation on methane uptake of a temperate forest 456 soil is influenced by labile carbon. Biology and Fertility of Soils 48, 621–631. 457 https://doi.org/10.1007/s00374-011-0660-3 458 Fest, B., Hinko-Najera, N., von Fischer, J.C., Livesley, S.J., Arndt, S.K., 2017. Soil 459 Methane Uptake Increases under Continuous Throughfall Reduction in a 460 Temperate Evergreen, Broadleaved Eucalypt Forest. Ecosystems 20, 368–379. 461 https://doi.org/10.1007/s10021-016-0030-y 462 Gundersen, P., Christiansen, J.R., Alberti, G., Brüggemann, N., Castaldi, S., Gasche, R., 463 Kitzler, B., Klemedtsson, L., Lobo-Do-Vale, R., Moldan, F., Rütting, T., Schleppi, 464 P., Weslien, P., Zechmeister-Boltenstern, S., 2012. The response of methane and 465 nitrous oxide fluxes to forest change in Europe. Biogeosciences 9, 3999–4012. 466 https://doi.org/10.5194/bg-9-3999-2012 467 Hartmann, A.A., Buchmann, N., Niklaus, P.A., 2011. A study of soil methane sink 468 regulation in two grasslands exposed to drought and N fertilization. Plant and Soil 469 342, 265–275. https://doi.org/10.1007/s11104-010-0690-x 470 Heimann, M., 2011. Enigma of the recent methane budget Bespoke cells for the human 471 brain. Nature 476, 157–158. 472 Hiltbrunner, D., Zimmermann, S., Karbin, S., Hagedorn, F., Niklaus, P.A., 2012. 473 Increasing soil methane sink along a 120-year afforestation chronosequence is 474 driven by soil moisture. Global Change Biology 18, 3664–3671. 475 https://doi.org/10.1111/j.1365-2486.2012.02798.x 25 476 Intergovernmental Panel on Climate Change (IPCC), 2007. Climate change 2007: the 477 physical science basis, Contribution of Working Group I to the Fourth Assessment 478 Report of the Intergovernmental Panel on Climate Change. Published for the 479 Intergovernmental Panel on Climate Change. 480 https://doi.org/10.1260/095830507781076194 481 Kai, F.M., Tyler, S.C., Randerson, J.T., Blake, D.R., 2011. Reduced methane growth 482 rate explained by decreased Northern Hemisphere microbial sources. Nature 476, 483 194–197. https://doi.org/10.1038/nature10259 484 Karbin, S., Hagedorn, F., Dawes, M.A., Niklaus, P.A., 2015. Treeline soil warming does 485 not affect soil methane fluxes and the spatial micro-distribution of methanotrophic 486 bacteria. Soil Biology and Biochemistry 86, 164–171. 487 https://doi.org/10.1016/j.soilbio.2015.03.022 488 King, G.M., 1997. Responses of atmospheric methane consumption by soils to global 489 climate change. Global Change Biology 3, 351–362. 490 https://doi.org/doi:10.1046/j.1365-2486.1997.00090.x 491 Kirschke, S., Bousquet, P., Ciais, P., Saunois, M., Canadell, J.G., Dlugokencky, E.J., 492 Bergamaschi, P., Bergmann, D., Blake, D.R., Bruhwiler, L., Cameron-Smith, P., 493 Castaldi, S., Chevallier, F., Feng, L., Fraser, A., Heimann, M., Hodson, E.L., 494 Houweling, S., Josse, B., Fraser, P.J., Krummel, P.B., Lamarque, J.F., Langenfelds, 495 R.L., Le Quéré, C., Naik, V., O’doherty, S., Palmer, P.I., Pison, I., Plummer, D., 496 Poulter, B., Prinn, R.G., Rigby, M., Ringeval, B., Santini, M., Schmidt, M., 497 Shindell, D.T., Simpson, I.J., Spahni, R., Steele, L.P., Strode, S.A., Sudo, K., 26 498 Szopa, S., Van Der Werf, G.R., Voulgarakis, A., Van Weele, M., Weiss, R.F., 499 Williams, J.E., Zeng, G., 2013. Three decades of global methane sources and sinks. 500 Nature Geoscience 6, 813–823. https://doi.org/10.1038/ngeo1955 501 Lin, X., Wang, S., Hu, Y., Luo, C., Zhang, Z., Niu, H., Xie, Z., 2015. Experimental 502 Warming Increases Seasonal Methane Uptake in an Alpine Meadow on the Tibetan 503 Plateau. Ecosystems 18, 274–286. https://doi.org/10.1007/s10021-014-9828-7 504 Lüke, C., Frenzel, P., 2011. Potential of pmoA amplicon pyrosequencing for 505 methanotroph diversity studies. Applied and Environmental Microbiology 77, 506 6305–6309. https://doi.org/10.1128/AEM.05355-11 507 Luo, G.J., Kiese, R., Wolf, B., Butterbach-Bahl, K., 2013. Effects of soil temperature 508 and moisture on methane uptake and nitrous oxide emissions across three different 509 ecosystem types. Biogeosciences 10, 3205–3219. https://doi.org/10.5194/bg-10- 510 3205-2013 511 Luo, Y., Gerten, D., Le Maire, G., Parton, W.J., WENG, E., ZHOU, X., KEOUGH, C., 512 BEIER, C., CIAIS, P., CRAMER, W., DUKES, J.S., EMMETT, B., HANSON, 513 P.J., KNAPP, A., LINDER, S., NEPSTAD, D., RUSTAD, L., 2008. Modeled 514 interactive effects of precipitation, temperature, and CO2 on ecosystem carbon and 515 water dynamics in different climatic zones. Global Change Biology 14, 1986– 516 1999. https://doi.org/10.1111/j.1365-2486.2008.01629.x 517 Matson, A., Pennock, D., Bedard-Haughn, A., 2009. Methane and nitrous oxide 518 emissions from mature forest stands in the boreal forest, Saskatchewan, Canada. 519 Forest Ecology and Management 258, 1073–1083. 27 520 https://doi.org/10.1016/j.foreco.2009.05.034 521 McLain, J.E.T., Martens, D. a, 2005. Studies of methane fluxes reveal that desert soils 522 can mitigate global climate change. Connecting mountain islands and desert seas: 523 biodiversity and management of the Madrean Archipelago II 496–499. 524 Mochizuki, Y., Koba, K., Yoh, M., 2012. Strong inhibitory effect of nitrate on 525 atmospheric methane oxidation in forest soils. Soil Biology and Biochemistry 50, 526 164–166. https://doi.org/10.1016/j.soilbio.2012.03.013 527 Nazaries, L., Tate, K.R., Ross, D.J., Singh, J., Dando, J., Saggar, S., Baggs, E.M., 528 Millard, P., Murrell, J.C., Singh, B.K., 2011. Response of methanotrophic 529 communities to afforestation and reforestation in New Zealand. ISME Journal 5, 530 1832–1836. https://doi.org/10.1038/ismej.2011.62 531 Ni, X., Groffman, P.M., 2018. Declines in methane uptake in forest soils. Proceedings 532 of the National Academy of Sciences 115, 201807377. 533 https://doi.org/10.1073/pnas.1807377115 534 Norby, R.J., Luo, Y., 2004. Evaluating ecosystem responses to rising atmospheric CO2 535 and global warming in a multi-factor world. New Phytologist 162, 281–293. 536 https://doi.org/10.1111/j.1469-8137.2004.01047.x 537 Pan, Y., Birdsey, R.A., Phillips, O.L., Jackson, R.B., 2013. The Structure, Distribution, 538 and Biomass of the World’s Forests. Annual Review of Ecology, Evolution, and 539 Systematics 44, 593–622. https://doi.org/10.1146/annurev-ecolsys-110512- 540 135914 541 Peñuelas, J., Poulter, B., Sardans, J., Ciais, P., Van Der Velde, M., Bopp, L., Boucher, 28 542 O., Godderis, Y., Hinsinger, P., Llusia, J., Nardin, E., Vicca, S., Obersteiner, M., 543 Janssens, I.A., 2013. Human-induced nitrogen-phosphorus imbalances alter 544 natural and managed ecosystems across the globe. Nature Communications 4. 545 https://doi.org/10.1038/ncomms3934 546 Phillips, R.L., Whalen, S.C., Schlesinger, W.H., 2001. Influence of atmospheric CO2 547 enrichment on methane consumption in a temperate forest ecosystem. Global 548 Change Biology 15, 741–752. 549 Price, S.J., Kelliher, F.M., Sherlock, R.R., Tate, K.R., Condron, L.M., 2004. 550 Environmental and chemical factors regulating methane oxidation in a New 551 Zealand forest soil. Australian Journal of Soil Research. 552 https://doi.org/10.1071/SR04026 553 Price, S.J., Sherlock, R.R., Kelliher, F.M., McSeveny, T.M., Tate, K.R., Condron, L.M., 554 2003. Pristine New Zealand forest soil is a strong methane sink. Global Change 555 Biology 10, 16–26. https://doi.org/10.1046/j.1529-8817.2003.00710x 556 Schlesinger, W.H., Bernhardt, E.S., 2013. Biogeochemistry: An Analysis of Global 557 Change, Third Edition, Biogeochemistry: An Analysis of Global Change, Third 558 Edition. https://doi.org/10.1016/C2010-0-66291-2 559 Smith, K.A., Dobbie, K.E., Ball, B.C., Bakken, L.R., Sitaula, B.K., Hansen, S., 560 Brumme, R., Borken, W., Christensen, S., Prieme, A., Fowler, D., MacDonald, 561 J.A., Skiba, U., Klemedtsson, L., Kasimir-Klemedtsson, A., Degorska, A., 562 Orlanski, P., 2000. Oxidation of atmosheric methane in Northen European soils, 563 comparison with other ecosystems, and uncertainties in the global terrestral sink. 29 564 Global Change Biology 6, 791–803. 565 Steinkamp, R., Butterbach-Bahl, K., Papen, H., 2001. Methane oxidation by soils of a 566 N-limited and N-fertilized spruce forest in the Black Forest, Germany. Soil 567 Biology & Biochemistry 33, 145–153. 568 Stephenson, N., 1990. climatic control of vegetation distribution: the role of the water 569 balance. American Naturalist. 570 Teh, Y.A., Diem, T., Jones, S., Huaraca Quispe, L.P., Baggs, E., Morley, N., Richards, 571 M., Smith, P., Meir, P., 2014. Methane and nitrous oxide fluxes across an elevation 572 gradient in the tropical Peruvian Andes. Biogeosciences 11, 2325–2339. 573 https://doi.org/10.5194/bg-11-2325-2014 574 Verchot, L. V., Davidson, E.A., Cattânio, J.H., Ackerman, I.L., 2000. Land-use change 575 and biogeochemical controls of methane fluxes in soils of eastern Amazonia. 576 Ecosystems 3, 41–56. https://doi.org/10.1007/s100210000009 577 Wang, Z.P., Ineson, P., 2003. Methane oxidation in a temperate coniferous forest soil: 578 Effects of inorganic N. Soil Biology and Biochemistry 35, 427–433. 579 https://doi.org/10.1016/S0038-0717(02)00294-8 580 Xu, X.F., Tian, H.Q., Zhang, C., Liu, M.L., Ren, W., Chen, G.S., Lu, C.Q., Bruhwiler, 581 L., 2010. Attribution of spatial and temporal variations in terrestrial methane flux 582 over North America. Biogeosciences 7, 3637–3655. https://doi.org/10.5194/bg-7- 583 3637-2010 584 Yu, L., Huang, Y., Zhang, W., Li, T., Sun, W., 2017. Methane uptake in global forest 585 and grassland soils from 1981 to 2010. Science of the Total Environment 607–608, 30 586 1163–1172. https://doi.org/10.1016/j.scitotenv.2017.07.082 587 31 588 Figure 1. Boxplots of the rate of methane uptake in soils of tropical (n=30), temperate 589 (n=89), and boreal (n=15) forests. Short dashes represent means, solid lines represent 590 medians, and error bars indicate the 90th and 10th percentiles. An extreme outlier at - 591 120 kg CH4 ha-1 y-1 in boreal forests has been omitted. 592 593 594 32 595 Figure 2. Partial residual plot. Partial residual plot of the variability of methane 596 uptake explained by MAT (°C), WS (water surplus (mm)) and the interaction between 597 them. 2a) the relationship between methane uptake and MAT under low, medium and 598 high WS (78, 405, 1010 mm); 2b) the relationship between methane uptake and WS 599 under low, medium and high MAT (4.5, 9.1, 21.8 °C) (visreg R package). 600 a Methane uptake (kg ha-1 y-1) b 33 601 Figure 3. Effect of the environmental-change components drought, warming, elevated 602 CO2, and N enrichment on the rates of methane uptake represented as percent of 603 change of the treatment versus the control values (horizontal bars, lower x-axis) and 604 the natural-log response ratios associated with each treatment (filled circles). * p < 605 0.05, *** p < 0.001, ns not significant. Numbers in brackets after the environmental- 606 change components indicate the number of experiments reviewed for each 607 component. 608 609 34